STPyV8

STPyV8 allows interoperability between Python 3 and JavaScript running Google V8 engine. Use STPyV8 to embed JavaScript code directly into your Python project, or to call Python code from JavaScript.

STPyV8 is a fork of the original PyV8 project, with code changed to work with the latest Google V8 engine and Python 3. STPyV8 links with Google V8 built as a static library.

Currently the library builds on Linux and MacOS, with Windows planned for the future.

Usage Examples

Wrapping a JavaScript function in a Python function:

# simple.py
import STPyV8

with STPyV8.JSContext() as ctxt:
  upcase = ctxt.eval("""
    ( (lowerString) => {
        return lowerString.toUpperCase();
    })
  """)
  print(upcase("hello world!"))

$ python simple.py
HELLO WORLD!

Using Python in V8

STPyV8 allows you to use Python functions, classes, and objects from within V8.

Exporting a Python class into V8 and using it from JavaScript:

# meaning.py
import STPyV8

class MyClass(STPyV8.JSClass):
  def reallyComplexFunction(self, addme):
    return 10 * 3 + addme

my_class = MyClass()

with STPyV8.JSContext(my_class) as ctxt:
  meaning = ctxt.eval("this.reallyComplexFunction(2) + 10;")
  print("The meaning of life: " + str(meaning))

$ python meaning.py
The meaning of life: 42

Using JavaScript in Python

STPyV8 allows you to use JavaScript functions, classes, object from Python.

Calling methods on a JavaScript class from Python code:

# circle.py
import STPyV8

with STPyV8.JSContext() as ctxt:
  ctxt.eval("""
    class Circle {
      constructor(radius) {
        this.radius = radius;
      }
      get area() {
        return this.calcArea()
      }
      calcArea() {
        return 3.14 * this.radius * this.radius;
      }
  }
  """)
  circle = ctxt.eval("new Circle(10)")
  print("Area of the circle: " + str(circle.area))

$ python cicle.py
Area of the circle: 314

Find more in the tests directory.

Installing

Installing on Linux/MacOS/Windows

python3 -m pip install st-pyv8

If no wheels are provided for your platform and Python version you are required to build STPyV8.

Building

GCC/clang or equivalent and Python3 headers are needed to build the main STPyV8 source code, as well as boost-python and some other boost dependencies.

Build Examples

Linux

python_version=3.7
echo Build with python $python_version

dirpath=$(dirname $(realpath $0))

cd docker

sudo docker build -t stpyv8-$python_version --build-arg ARG_PYTHON_VERSION=$python_version .

cd ../

sudo docker run -it -v $dirpath:/stpyv8 --entrypoint /bin/bash stpyv8-$python_version -c 'cd /stpyv8 && python setup.py bdist_wheel --plat-name=manylinux1_x86_64'

MacOS

Building on MacOS requires full XCode (not just the command line tools) to compile Google V8. The command line tools bundled with XCode are required (rather than the stand-alone command line tools, sometimes requiring drastic measures .)

export PATH=/usr/local/Cellar/python@3.10/3.10.6_2/bin:$PATH
ln -s /usr/local/Cellar/python@3.10/3.10.6_2/bin/python3 /usr/local/Cellar/python@3.10/3.10.6_2/bin/python
# ensure default python point to the target python path
python setup.py bdist_wheel --plat-name macosx_10_10_x86_64

Windows

Building on Windows requires Microsoft's development tools to be installed

set PATH=C:\Users\shadowyang\AppData\Local\Programs\Python\Python310;%PATH%
:: ensure default python point to the target python path
python setup.py bdist_wheel

More detailed build instructions are in the docs folder.

How does this work?

STPyV8 is a Python C++ Extension Module that links to an embedded V8 library. Since PyV8 used the Boost.Python C++ library (as wells as some others) we kept it, but future work may include just using the C API exposed by Python and eliminating boost. Think of this as an Oreo cookie - Python and Google V8 crackers with C++ icing in the middle gluing them together.

Is STPyV8 fast?

STPyV8 needs to translate Python arguments (and JavaScript arguments) back and forth between function and method calls in the two languages. It does the minimum amount of work using native code, but if you are interested in the height of performance, make your interface between Python and JavaScript "chunky" ... i.e., make the minimum number of transitions between the two.

What can I use this for?

We use STPyV8 to simulate a browser, and then execute sketchy JavaScript in an instrumented container. Other kinds of JavaScript sandboxing (simulating and monitoring the external world to JavaScript code) are a natural fit.

gongqf/stpyv8